PyTorchKR Papers

최신 논문

총 923편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2502.01673공식 코드 · Papers with Code

Multilingual State Space Models for Structured Question Answering in Indic Languages

The diversity and complexity of Indic languages present unique challenges for natural language processing (NLP) tasks, particularly in the domain of question answering (QA).To address these challenges, this paper explores the application of State Space Models (SSMs),to build efficient and contextually aware QA systems tailored for Indic languages. SSMs are particularly suited for this task due to their ability to model long-term and short-term dependencies in sequential data, making them well-equipped to handle the rich morphology, complex syntax, and contextual intricacies characteristic of Indian languages. We evaluated multiple SSM architectures across diverse datasets representing various Indic languages and conducted a comparative analysis of their performance. Our results demonstrate that these models effectively capture linguistic subtleties, leading to significant improvements in question interpretation, context alignment, and answer generation. This work represents the first application of SSMs to question answering tasks in Indic languages, establishing a foundational benchmark for future research in this domain. We propose enhancements to existing SSM frameworks, optimizing their applicability to low-resource settings and multilingual scenarios prevalent in Indic languages.

arXiv:2502.00436

Secure Data Reconstruction: A Direct Data-Driven Approach

This paper addresses the problem of secure data reconstruction for unknown systems, where data collected from the system are susceptible to malicious manipulation. We aim to recover the real trajectory without prior knowledge of the system model. To achieve this, a behavioral language is used to represent the system, describing it using input/output trajectories instead of state-space models. We consider two attack scenarios. In the first scenario, up to $k$ entries of the collected data are malicious. On the other hand, the second scenario assumes that at most $k$ channels from sensors or actuators can be compromised, implying that any data collected from these channels might be falsified. For both scenarios, we formulate the trajectory recovery problem as an optimization problem and introduce sufficient conditions to ensure successful recovery of the true data. Since finding exact solutions to these problems can be computationally inefficient, we further approximate them using an $\ell_1$-norm and group Least Absolute Shrinkage and Selection Operator (LASSO). We demonstrate that under certain conditions, these approximation problems also find the true trajectory while maintaining low computation complexity. Finally, we extend the proposed algorithms to noisy data. By reconstructing the secure trajectory, this work serves as a safeguard mechanism for subsequent data-driven control methods.

arXiv:2502.00435공식 코드 · Papers with Code

SatMamba: Development of Foundation Models for Remote Sensing Imagery Using State Space Models

Foundation models refer to deep learning models pretrained on large unlabeled datasets through self-supervised algorithms. In the Earth science and remote sensing communities, there is growing interest in transforming the use of Earth observation data, including satellite and aerial imagery, through foundation models. Various foundation models have been developed for remote sensing, such as those for multispectral, high-resolution, and hyperspectral images, and have demonstrated superior performance on various downstream tasks compared to traditional supervised models. These models are evolving rapidly, with capabilities to handle multispectral, multitemporal, and multisensor data. Most studies use masked autoencoders in combination with Vision Transformers (ViTs) as the backbone for pretraining. While the models showed promising performance, ViTs face challenges, such as quadratic computational scaling with input length, which may limit performance on multiband and multitemporal data with long sequences. This research aims to address these challenges by proposing SatMamba, a new pretraining framework that combines masked autoencoders with State Space Model, offering linear computational scaling. Experiments on high-resolution imagery across various downstream tasks show promising results, paving the way for more efficient foundation models and unlocking the full potential of Earth observation data. The source code is available in https://github.com/mdchuc/HRSFM.

arXiv:2501.16639

Finite Sample Analysis of Subspace Identification Methods

As one of the mainstream approaches in system identification, subspace identification methods (SIMs) are known for their simple parameterization for MIMO systems and robust numerical properties. However, a comprehensive statistical analysis of SIMs remains an open problem. Amid renewed focus on identifying state-space models in the non-asymptotic regime, this work presents a finite sample analysis for a large class of open-loop SIMs. It establishes high-probability upper bounds for system matrices obtained via SIMs, and reveals that convergence rates for estimating Markov parameters and system matrices are $\mathcal{O}(1/\sqrt{N})$ up to logarithmic terms, in line with classical asymptotic results. Following the key steps of SIMs, we arrive at the above results by a three-step procedure. In Step 1, we begin with a parsimonious SIM (PARSIM) that uses least-squares regression to estimate multiple high-order ARX models in parallel. Leveraging a recent analysis of an individual ARX model, we obtain a union error bound for a bank of ARX models. Step 2 involves model reduction via weighted singular value decomposition (SVD), where we consider different data-dependent weighting matrices and use robustness results for SVD to obtain error bounds on extended controllability and observability matrices, respectively. The final Step 3 focuses on deriving error bounds for system matrices, where two different realization algorithms, the MOESP type and the Larimore type, are considered. Although our study initially focuses on PARSIM, the methodologies apply broadly across many variants of SIMs.

arXiv:2501.16384

MambaTron: Efficient Cross-Modal Point Cloud Enhancement using Aggregate Selective State Space Modeling

Point cloud enhancement is the process of generating a high-quality point cloud from an incomplete input. This is done by filling in the missing details from a reference like the ground truth via regression, for example. In addition to unimodal image and point cloud reconstruction, we focus on the task of view-guided point cloud completion, where we gather the missing information from an image, which represents a view of the point cloud and use it to generate the output point cloud. With the recent research efforts surrounding state-space models, originally in natural language processing and now in 2D and 3D vision, Mamba has shown promising results as an efficient alternative to the self-attention mechanism. However, there is limited research towards employing Mamba for cross-attention between the image and the input point cloud, which is crucial in multi-modal problems. In this paper, we introduce MambaTron, a Mamba-Transformer cell that serves as a building block for our network which is capable of unimodal and cross-modal reconstruction which includes view-guided point cloud completion.We explore the benefits of Mamba's long-sequence efficiency coupled with the Transformer's excellent analytical capabilities through MambaTron. This approach is one of the first attempts to implement a Mamba-based analogue of cross-attention, especially in computer vision. Our model demonstrates a degree of performance comparable to the current state-of-the-art techniques while using a fraction of the computation resources.

arXiv:2501.17088공식 코드 · Papers with Code

Mamba-Shedder: Post-Transformer Compression for Efficient Selective Structured State Space Models

Large pre-trained models have achieved outstanding results in sequence modeling. The Transformer block and its attention mechanism have been the main drivers of the success of these models. Recently, alternative architectures, such as Selective Structured State Space Models (SSMs), have been proposed to address the inefficiencies of Transformers. This paper explores the compression of SSM-based models, particularly Mamba and its hybrids. We study the sensitivity of these models to the removal of selected components at different granularities to reduce the model size and computational overhead, thus improving their efficiency while maintaining accuracy. The proposed solutions, collectively referred to as Mamba-Shedder, achieve a speedup of up to 1.4x during inference, demonstrating that model efficiency can be improved by eliminating several redundancies with minimal impact on the overall model performance. The code is available at https://github.com/IntelLabs/Hardware-Aware-Automated-Machine-Learning.

arXiv:2501.16237

Application of Structured State Space Models to High energy physics with locality-sensitive hashing

Modern high-energy physics (HEP) experiments are increasingly challenged by the vast size and complexity of their datasets, particularly regarding large-scale point cloud processing and long sequences. In this study, to address these challenges, we explore the application of structured state space models (SSMs), proposing one of the first trials to integrate local-sensitive hashing into either a hybrid or pure Mamba Model. Our results demonstrate that pure SSMs could serve as powerful backbones for HEP problems involving tasks for long sequence data with local inductive bias. By integrating locality-sensitive hashing into Mamba blocks, we achieve significant improvements over traditional backbones in key HEP tasks, surpassing them in inference speed and physics metrics while reducing computational overhead. In key tests, our approach demonstrated promising results, presenting a viable alternative to traditional transformer backbones by significantly reducing FLOPS while maintaining robust performance.

arXiv:2501.14850

On the locality bias and results in the Long Range Arena

The Long Range Arena (LRA) benchmark was designed to evaluate the performance of Transformer improvements and alternatives in long-range dependency modeling tasks. The Transformer and its main variants performed poorly on this benchmark, and a new series of architectures such as State Space Models (SSMs) gained some traction, greatly outperforming Transformers in the LRA. Recent work has shown that with a denoising pre-training phase, Transformers can achieve competitive results in the LRA with these new architectures. In this work, we discuss and explain the superiority of architectures such as MEGA and SSMs in the Long Range Arena, as well as the recent improvement in the results of Transformers, pointing to the positional and local nature of the tasks. We show that while the LRA is a benchmark for long-range dependency modeling, in reality most of the performance comes from short-range dependencies. Using training techniques to mitigate data inefficiency, Transformers are able to reach state-of-the-art performance with proper positional encoding. In addition, with the same techniques, we were able to remove all restrictions from SSM convolutional kernels and learn fully parameterized convolutions without decreasing performance, suggesting that the design choices behind SSMs simply added inductive biases and learning efficiency for these particular tasks. Our insights indicate that LRA results should be interpreted with caution and call for a redesign of the benchmark.

arXiv:2501.16295공식 코드 · Papers with Code

Mixture-of-Mamba: Enhancing Multi-Modal State-Space Models with Modality-Aware Sparsity

State Space Models (SSMs) have emerged as efficient alternatives to Transformers for sequential modeling, but their inability to leverage modality-specific features limits their performance in multi-modal pretraining. Here, we propose Mixture-of-Mamba, a novel SSM architecture that introduces modality-aware sparsity through modality-specific parameterization of the Mamba block. Building on Mixture-of-Transformers (W. Liang et al. arXiv:2411.04996; 2024), we extend the benefits of modality-aware sparsity to SSMs while preserving their computational efficiency. We evaluate Mixture-of-Mamba across three multi-modal pretraining settings: Transfusion (interleaved text and continuous image tokens with diffusion loss), Chameleon (interleaved text and discrete image tokens), and an extended three-modality framework incorporating speech. Mixture-of-Mamba consistently reaches the same loss values at earlier training steps with significantly reduced computational costs. In the Transfusion setting, Mixture-of-Mamba achieves equivalent image loss using only 34.76% of the training FLOPs at the 1.4B scale. In the Chameleon setting, Mixture-of-Mamba reaches similar image loss with just 42.50% of the FLOPs at the 1.4B scale, and similar text loss with just 65.40% of the FLOPs. In the three-modality setting, MoM matches speech loss at 24.80% of the FLOPs at the 1.4B scale. Our ablation study highlights the synergistic effects of decoupling projection components, where joint decoupling yields greater gains than individual modifications. These results establish modality-aware sparsity as a versatile and effective design principle, extending its impact from Transformers to SSMs and setting new benchmarks in multi-modal pretraining. Our code can be accessed at https://github.com/Weixin-Liang/Mixture-of-Mamba

arXiv:2501.14673

State Space Models for Extractive Summarization in Low Resource Scenarios

Extractive summarization involves selecting the most relevant sentences from a text. Recently, researchers have focused on advancing methods to improve state-of-the-art results in low-resource settings. Motivated by these advancements, we propose the MPoincareSum method. This method applies the Mamba state space model to generate the semantics of reviews and sentences, which are then concatenated. A Poincare compression is used to select the most meaningful features, followed by the application of a linear layer to predict sentence relevance based on the corresponding review. Finally, we paraphrase the relevant sentences to create the final summary. To evaluate the effectiveness of MPoincareSum, we conducted extensive experiments using the Amazon review dataset. The performance of the method was assessed using ROUGE scores. The experimental results demonstrate that MPoincareSum outperforms several existing approaches in the literature

arXiv:2501.13624

QMamba: Post-Training Quantization for Vision State Space Models

State Space Models (SSMs), as key components of Mamaba, have gained increasing attention for vision models recently, thanks to their efficient long sequence modeling capability. Given the computational cost of deploying SSMs on resource-limited edge devices, Post-Training Quantization (PTQ) is a technique with the potential for efficient deployment of SSMs. In this work, we propose QMamba, one of the first PTQ frameworks to our knowledge, designed for vision SSMs based on the analysis of the activation distributions in SSMs. We reveal that the distribution of discrete parameters exhibits long-tailed skewness and the distribution of the hidden state sequence exhibits highly dynamic variations. Correspondingly, we design Long-tailed Skewness Quantization (LtSQ) to quantize discrete parameters and Temporal Group Quantization (TGQ) to quantize hidden states, which reduces the quantization errors. Extensive experiments demonstrate that QMamba outperforms advanced PTQ methods on vision models across multiple model sizes and architectures. Notably, QMamba surpasses existing methods by 21.0% on ImageNet classification with 4-bit activations.

arXiv:2501.13353

Contrast: A Hybrid Architecture of Transformers and State Space Models for Low-Level Vision

Transformers have become increasingly popular for image super-resolution (SR) tasks due to their strong global context modeling capabilities. However, their quadratic computational complexity necessitates the use of window-based attention mechanisms, which restricts the receptive field and limits effective context expansion. Recently, the Mamba architecture has emerged as a promising alternative with linear computational complexity, allowing it to avoid window mechanisms and maintain a large receptive field. Nevertheless, Mamba faces challenges in handling long-context dependencies when high pixel-level precision is required, as in SR tasks. This is due to its hidden state mechanism, which can compress and store a substantial amount of context but only in an approximate manner, leading to inaccuracies that transformers do not suffer from. In this paper, we propose \textbf{Contrast}, a hybrid SR model that combines \textbf{Con}volutional, \textbf{Tra}nsformer, and \textbf{St}ate Space components, effectively blending the strengths of transformers and Mamba to address their individual limitations. By integrating transformer and state space mechanisms, \textbf{Contrast} compensates for the shortcomings of each approach, enhancing both global context modeling and pixel-level accuracy. We demonstrate that combining these two architectures allows us to mitigate the problems inherent in each, resulting in improved performance on image super-resolution tasks.

arXiv:2501.12732

GRAMA: Adaptive Graph Autoregressive Moving Average Models

Graph State Space Models (SSMs) have recently been introduced to enhance Graph Neural Networks (GNNs) in modeling long-range interactions. Despite their success, existing methods either compromise on permutation equivariance or limit their focus to pairwise interactions rather than sequences. Building on the connection between Autoregressive Moving Average (ARMA) and SSM, in this paper, we introduce GRAMA, a Graph Adaptive method based on a learnable Autoregressive Moving Average (ARMA) framework that addresses these limitations. By transforming from static to sequential graph data, GRAMA leverages the strengths of the ARMA framework, while preserving permutation equivariance. Moreover, GRAMA incorporates a selective attention mechanism for dynamic learning of ARMA coefficients, enabling efficient and flexible long-range information propagation. We also establish theoretical connections between GRAMA and Selective SSMs, providing insights into its ability to capture long-range dependencies. Extensive experiments on 14 synthetic and real-world datasets demonstrate that GRAMA consistently outperforms backbone models and performs competitively with state-of-the-art methods.

arXiv:2501.12215

Automatic selection of the best neural architecture for time series forecasting via multi-objective optimization and Pareto optimality conditions

Time series forecasting plays a pivotal role in a wide range of applications, including weather prediction, healthcare, structural health monitoring, predictive maintenance, energy systems, and financial markets. While models such as LSTM, GRU, Transformers, and State-Space Models (SSMs) have become standard tools in this domain, selecting the optimal architecture remains a challenge. Performance comparisons often depend on evaluation metrics and the datasets under analysis, making the choice of a universally optimal model controversial. In this work, we introduce a flexible automated framework for time series forecasting that systematically designs and evaluates diverse network architectures by integrating LSTM, GRU, multi-head Attention, and SSM blocks. Using a multi-objective optimization approach, our framework determines the number, sequence, and combination of blocks to align with specific requirements and evaluation objectives. From the resulting Pareto-optimal architectures, the best model for a given context is selected via a user-defined preference function. We validate our framework across four distinct real-world applications. Results show that a single-layer GRU or LSTM is usually optimal when minimizing training time alone. However, when maximizing accuracy or balancing multiple objectives, the best architectures are often composite designs incorporating multiple block types in specific configurations. By employing a weighted preference function, users can resolve trade-offs between objectives, revealing novel, context-specific optimal architectures. Our findings underscore that no single neural architecture is universally optimal for time series forecasting. Instead, the best-performing model emerges as a data-driven composite architecture tailored to user-defined criteria and evaluation objectives.

arXiv:2501.12381

Parallel Sequence Modeling via Generalized Spatial Propagation Network

We present the Generalized Spatial Propagation Network (GSPN), a new attention mechanism optimized for vision tasks that inherently captures 2D spatial structures. Existing attention models, including transformers, linear attention, and state-space models like Mamba, process multi-dimensional data as 1D sequences, compromising spatial coherence and efficiency. GSPN overcomes these limitations by directly operating on spatially coherent image data and forming dense pairwise connections through a line-scan approach. Central to GSPN is the Stability-Context Condition, which ensures stable, context-aware propagation across 2D sequences and reduces the effective sequence length to $\sqrt{N}$ for a square map with N elements, significantly enhancing computational efficiency. With learnable, input-dependent weights and no reliance on positional embeddings, GSPN achieves superior spatial fidelity and state-of-the-art performance in vision tasks, including ImageNet classification, class-guided image generation, and text-to-image generation. Notably, GSPN accelerates SD-XL with softmax-attention by over $84\times$ when generating 16K images.

arXiv:2501.11729

SeRpEnt: Selective Resampling for Expressive State Space Models

State Space Models (SSMs) have recently enjoyed a rise to prominence in the field of deep learning for sequence modeling, especially as an alternative to Transformers. Their success stems from avoiding two well-known drawbacks of attention-based models: quadratic complexity with respect to the sequence length and inability to model long-range dependencies. The SSM variant Mamba has demonstrated performance comparable to Transformers without any form of attention, thanks to the use of a selective mechanism for the state parameters. Selectivity, however, is only evaluated empirically and the reasons of its effectiveness remain unclear. In this work, we show how selectivity is related to the sequence processing. Our analysis shows that selective time intervals in Mamba act as linear approximators of information. Then, we propose our SeRpEnt architecture, a SSM that further exploits selectivity to compress sequences in an information-aware fashion. It employs a resampling mechanism that aggregates elements based on their information content. Our empirical results in the Long Range Arena benchmark and other language modeling tasks show benefits of the SeRpEnt's resampling mechanism.

arXiv:2501.11357

On the dimension of pullback attractors in recurrent neural networks

Recurrent Neural Networks (RNNs) are high-dimensional state space models capable of learning functions on sequence data. Recently, it has been conjectured that reservoir computers, a particular class of RNNs, trained on observations of a dynamical systems can be interpreted as embeddings. This result has been established for the case of linear reservoir systems. In this work, we use a nonautonomous dynamical systems approach to establish an upper bound for the fractal dimension of the subset of reservoir state space approximated during training and prediction phase. We prove that when the input sequences comes from an Nin-dimensional invertible dynamical system, the fractal dimension of this set is bounded above by Nin. The result obtained here are useful in dimensionality reduction of computation in RNNs as well as estimating fractal dimensions of dynamical systems from limited observations of their time series. It is also a step towards understanding embedding properties of reservoir computers.

arXiv:2501.11238

WSSM: Geographic-enhanced hierarchical state-space model for global station weather forecast

Global Station Weather Forecasting (GSWF), a prominent meteorological research area, is pivotal in providing timely localized weather predictions. Despite the progress existing models have made in the overall accuracy of the GSWF, executing high-precision extreme event prediction still presents a substantial challenge. The recent emergence of state-space models, with their ability to efficiently capture continuous-time dynamics and latent states, offer potential solutions. However, early investigations indicated that Mamba underperforms in the context of GSWF, suggesting further adaptation and optimization. To tackle this problem, in this paper, we introduce Weather State-space Model (WSSM), a novel Mamba-based approach tailored for GSWF. Geographical knowledge is integrated in addition to the widely-used positional encoding to represent the absolute special-temporal position. The multi-scale time-frequency features are synthesized from coarse to fine to model the seasonal to extreme weather dynamic. Our method effectively improves the overall prediction accuracy and addresses the challenge of forecasting extreme weather events. The state-of-the-art results obtained on the Weather-5K subset underscore the efficacy of the WSSM

arXiv:2501.12352

Test-time regression: a unifying framework for designing sequence models with associative memory

Sequence models lie at the heart of modern deep learning. However, rapid advancements have produced a diversity of seemingly unrelated architectures, such as Transformers and recurrent alternatives. In this paper, we introduce a unifying framework to understand and derive these sequence models, inspired by the empirical importance of associative recall, the capability to retrieve contextually relevant tokens. We formalize associative recall as a two-step process, memorization and retrieval, casting memorization as a regression problem. Layers that combine these two steps perform associative recall via ``test-time regression'' over its input tokens. Prominent layers, including linear attention, state-space models, fast-weight programmers, online learners, and softmax attention, arise as special cases defined by three design choices: the regression weights, the regressor function class, and the test-time optimization algorithm. Our approach clarifies how linear attention fails to capture inter-token correlations and offers a mathematical justification for the empirical effectiveness of query-key normalization in softmax attention. Further, it illuminates unexplored regions within the design space, which we use to derive novel higher-order generalizations of softmax attention. Beyond unification, our work bridges sequence modeling with classic regression methods, a field with extensive literature, paving the way for developing more powerful and theoretically principled architectures.

arXiv:2501.06881

Gaussian Integral based Bayesian Smoother

This work introduces the Gaussian integration to address a smoothing problem of a nonlinear stochastic state space model. The probability densities of states at each time instant are assumed to be Gaussian, and their means and covariances are evaluated by utilizing the odd-even properties of Gaussian integral, which are further utilized to realize Rauch-Tung-Striebel (RTS) smoothing expressions. Given that the Gaussian integration provides an exact solution for the integral of a polynomial function over a Gaussian probability density function, it is anticipated to provide more accurate results than other existing Gaussian approximation-based smoothers such as extended Kalman, cubature Kalman, and unscented Kalman smoothers, especially when polynomial types of nonlinearity are present in the state space models. The developed smoothing algorithm is applied to the Van der Pol oscillator, where the nonlinearity associated with their dynamics is represented using polynomial functions. Simulation results are provided to demonstrate the superiority of the proposed algorithm.